LINKING MINERALOGY TO GASTROINTESTINAL BIOAVAILABILITY AND BIOACCESSIBILTY OF ARSENIC AT THE EMPIRE MINE STATE PARK, CA, USA

We determined the mineralogical hosts of arsenic (As) in > 20 sediment samples from the Empire Mine State Park (EMSP) in California and compared that data to in vivo gastrointestinal relative bioavailability (RBA) measured in juvenile swine and to As liberated by an in vitro, physiologically-based extraction test called the “California Arsenic Bioaccessibility” (CAB) method, which was developed as part of the study. The relationship between RBA and CAB is highly predictive and correlative for samples with less than about 1,500 mg/kg As (RBA = 0.82(CAB) + 2.92, r² = 0.91), but CAB predicts about twice the RBA value for samples with higher As concentrations.

We employed “bulk” chemical/mineralogical analyses including coupled (XRF) and XRD (in the USGS computer package “ROCKJOCK”), synchrotron-based X-ray absorption spectroscopy (XAS), and sequential chemical extraction data. Micron-scale analyses included electron microprobe (EM), scanning electron microscopy (SEM), the quantitative mineralogy technique known as QEMSCAN®, microRaman spectroscopy, and focused-beam synchrotron-based XRF and XAS.

Pearson correlations between the mineralogical datasets and either RBA or CAB were explored as a first step to identify the minerals hosting bioavailable As. This analysis suggests that As(V) associated with Fe(III) hydroxide (Fehydr) appears to be the main phase controlling As RBA and CAB in EMSP samples. Significant positive (p < 0.05) correlations with RBA and/or CAB were found for (1) the abundance of As(V)- relative to other As-hosting phases (bulk XAS); (2) the average amount of As in Fehydr (electron microprobe), and (3) the abundance of Fehydr itself (bulk XAS). Significant positive correlations were also found between RBA or CAB and the relative abundance of As (V and III) associated with Al oxyhydroxide, gibbsite, or kaolinite (bulk XAS). Relatively soluble, high As phases (e.g., scorodite, FeAsO₄ 2H₂O) were identified by micron-scale techniques, but did not provide statistically significant correlations due to low abundance in samples and limited distribution across samples. We are currently investigating the possibility that these phases are the cause of the extreme overprediction of the CAB method for samples with high total As.